Geolocation System Using Non-Parametric Bayesian Clustering

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional geolocation techniques fail to accurately determine the location of radio emitters in multi-path environments due to non-line-of-sight conditions and the presence of multiple emitters, as they rely on prior assumptions and are ineffective in discriminating correct emitter contributions, leading to inaccurate location tracking.

Innovation Solution

A non-parametric Bayesian clustering technique using a Dirichlet Process Mixture Model (DPMM) is employed to cluster Angle of Arrival (AOA) and Time of Arrival (TOA) data without prior assumptions, combined with a cognitive sensor activation framework to selectively activate sensors, optimizing power consumption and reducing noise.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional geolocation techniques using triangulation are used, then the system can determine emitter location, but the accuracy deteriorates in multi-path environments with non-line-of-sight conditions

Engineering Contradiction:

Improveemitter location accuracyVSAvoidgeolocation reliability in multi-path environment

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent changes the fundamental parameters used for geolocation from direct triangulation (AOA, TOA, TDOA, RSS, FDOA) to a clustering-based approach that groups multi-path components by their spatial and temporal characteristics. This parameter transformation allows the system to identify line-of-sight components among multiple paths, thereby maintaining accuracy in non-line-of-sight conditions.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent segments the received signal into multiple path components and clusters them based on AOA and TOA characteristics. By dividing the complex multi-path signal into distinguishable clusters, the system can identify and select the line-of-sight path for accurate geolocation, resolving the reliability issue in multi-path environments.

Inventive Principle:

Principle #1Segmentation

2Device complexity

If prior assumptions are made about the number of emitters, then the clustering process can be simplified, but the accuracy of emitter discrimination deteriorates

Engineering Contradiction:

Improveclustering process complexityVSAvoidemitter discrimination accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent employs a self-organizing clustering algorithm that automatically determines the number of emitter clusters based on the received signal characteristics without requiring prior assumptions. The algorithm autonomously identifies the correct number of clusters through data-driven analysis, eliminating the need for manual configuration while maintaining high discrimination accuracy.

Inventive Principle:

Principle #25Self-service

3Reliability

If all sensors are activated continuously, then complete coverage is maintained, but power consumption increases

Engineering Contradiction:

Improvesensor network coverageVSAvoidsensor network power consumption

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent implements periodic sensor activation based on detected emitter activity and location changes. Sensors are activated only when needed for tracking emitters in their current locations, rather than continuous operation. This periodic activation maintains necessary coverage while dramatically reducing overall network power consumption.

Inventive Principle:

Principle #19Periodic action

Solution Approach 2:

The patent applies local quality by activating only those sensors that are spatially relevant to the current emitter locations. Each sensor's activation state is optimized based on its local geographic relationship to emitters, ensuring adequate local coverage while minimizing unnecessary activation of distant sensors.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS9804253B2System and methods for non-parametric technique based geolocation and cognitive sensor activation

Publication Date: 2017.10.31 ROCKWELL COLLINS INC

AI summary

The present invention relates to a geolocation system and method for a multi-path environment. The geolocation system comprises one or more emitters (201a . . . 201n), one or more sensors (202a . . . 202n) comprising at least one processor. A first processor (204) estimates angle of arrival (AOA) and time of arrival (TOA) from the signals received from said one or more emitters (201a . . . 201n). A second processor (205) determines clusters based on the (AOA) and (TOA) data. The system also comprises a central node (207) in communication with at least one sensor (202a . . . 202n) and configured to estimate geolocation of one or more emitters (201a . . . 201n) wherein, said second processor (205) clusters data for the one or more emitters (201a . . . 201n) by executing a non-parametric Bayesian technique and said central node (207) utilizes hybrid angle of arrival-time difference of arrival (AOA-TDOA) technique to determine geolocation of each of the emitters (201a . . . 201n).